Leadbay wants to sell you the customers your database can't see

Ask any sales rep who has worked a territory of roofers, restaurant suppliers, or regional fabricators, and they will tell you the same thing: the best prospects are the ones you can't look up. They have a one-page website, if that. No press releases. No enrichment record. They exist on invoices and word of mouth. Leadbay, a Y Combinator company from the Fall 2025 batch, was built on the observation that this is not a small corner of the market. It is most of it.

The pitch is direct. Traditional prospecting tools - LinkedIn Sales Navigator, ZoomInfo, Apollo, Clay - are very good at surfacing companies that leave a big digital trail. That works for software firms and mid-market brands. It works poorly for the SMB economy, where the digital footprint is thin or fragmented. Leadbay's argument is that these businesses aren't missing. The tools looking for them are just built to see the wrong signals.

01 / THE PRODUCTA model, not a database

Instead of querying a static database, Leadbay trains a domain-specific AI model on three inputs: broad market data, a company's own internal knowledge (CRM, ERP, a CSV of past deals), and the harder-to-capture instinct of its reps. The model learns what a good customer looks like for that specific business, then goes looking for companies that match - including ones no database would have returned. Co-founder and CTO Milan Stankovic holds a PhD in exactly this problem: making AI reason usefully when the data is sparse.

The practical unit of value is the qualified lead. Leadbay learns a team's Ideal Customer Profile from its won and lost deals, scores new prospects by likelihood to buy, and answers qualification questions through what it calls signal crawling - reading scattered public traces the way an experienced rep would. A waterfall enrichment layer pulls from more than 20 databases to fill in contact and firmographic detail.

02 / THE DIFFERENCEReasoning where the data runs out

The clearest way to see Leadbay's position is to line it up against the tools reps already pay for. In markets with a rich digital footprint, everyone performs. The gap opens in the low-signal industries - construction, hospitality, manufacturing, regional B2B services - where database coverage thins out and Leadbay's reasoning approach is designed to keep going.

Qualified leads found in low-signal SMB markets (illustrative)

LinkedInLow

ZoomInfoLow

ApolloLow

ClayMed

LeadbayHigh

The other difference is delivery. Leadbay's bet is that reps shouldn't have to open one more dashboard. Through an MCP library of native components, it pushes qualified leads and sales tools into the AI apps people already use - Claude, ChatGPT, Copilot. The company puts it plainly: the next Salesforce, it argues, will be a set of sales and marketing blocks that live inside those assistants rather than a standalone suite.

03 / HOW IT LANDSProve the lift, then expand

For enterprise customers, Leadbay sells outcomes before software. A forward-deployed engineer integrates the platform with existing systems, tunes the model on real historical deals, and aims to demonstrate a roughly 5x prospecting improvement before the rollout widens across regions, teams, and segments under shared models and governance.

04 / WHO USES ITFrom roofers to L'Oreal

The customer base spans small teams working thin territories and enterprise brands that want to expand a defined addressable market. Named users include large industrial and consumer names alongside high-growth software firms - a spread that fits the claim that data-scarce prospecting is a horizontal problem, not a niche one.

The industries they sell into read like the parts of the economy that rarely trend online: construction, hospitality and furniture, manufacturing, wholesaling, staffing, retail, and broad B2B services. Pricing runs from a free tier of a handful of companies a week up through $100 and $400 per-user monthly plans, with custom enterprise deployments on top.

05 / THE MONEYA $4.3M seed, and a research bet

In May 2026, Leadbay closed a $4.3M seed round led by Y Combinator. The plan for the capital is unglamorous and specific: grow the U.S. go-to-market team in San Francisco, expand engineering, and fund a research partnership with Sorbonne University aimed at strengthening the model's inference on low-data problems - the academic version of the company's whole reason to exist.

• Y Combinator (lead)
• Rebel Ventures
• Roosh Ventures
• Inovexus Ventures
• TS Ventures
• Alumni Ventures
• Bright Ventures
• Transpose Platform
• Deel Ventures

06 / THE FOUNDERSA limited-data PhD and a community builder

The pairing is the point. Ludovic Granger, the CEO, was an early employee at an EU-US venture firm and built a community of some 300 sales reps before starting the company - useful when your product has to be trusted by exactly those people. Milan Stankovic, the CTO, brings prior exit experience and a doctorate in AI with limited data. One founder knows what reps actually do; the other knows how to make a model behave when the inputs are thin.

Founded in 2023 and now headquartered in San Francisco with roots in France, Leadbay is small - Y Combinator lists a team of around eight - and deliberately technical. The company frames its target not as a list of leads but as an entire "invisible SMB economy," and its work as teaching software to sell where the data trail goes cold.

Whether Leadbay becomes the block that lives inside every AI assistant, or one of several companies chasing that idea, is not yet decided. What is clear is the wager: that the most valuable prospects in B2B are the ones nobody else can find, and that a model - not a bigger database - is how you find them.